Explore the vast range of titles available.

Many research questions involve comparing predictions or effects across multiple models. For example, it may be of interest whether an independent variable's effect changes after adding variables to a model. Or, it could be important to compare a variable's effect on different outcomes or across different types of models. When doing this, marginal effects are a useful method for quantifying effects because they are in the natural metric of the dependent variable and they avoid identification problems when comparing regression coefficients across logit and probit models. Despite advances that make it possible to compute marginal effects for almost any model, there is no general method for comparing these effects across models. In this article, the authors provide a general framework for comparing predictions and marginal effects across models using seemingly unrelated estimation to combine estimates from multiple models, which allows tests of the equality of predictions and effects across models. The authors illustrate their method to compare nested models, to compare effects on different dependent or independent variables, to compare results from different samples or groups within one sample, and to assess results from different types of models.

Bioenergy is expected to play an important role in the achievement of stringent climate-change mitigation targets requiring the application of negative emissions technology. Using a multi-model framework, we assess the effects of high bioenergy demand on global food production, food security, and competition for agricultural land. Various scenarios simulate global bioenergy demands of 100, 200, 300, and 400 exajoules (EJ) by 2100, with and without a carbon price. Six global energy-economy-agriculture models contribute to this study, with different methodologies and technologies used for bioenergy supply and greenhouse-gas mitigation options for agriculture. We find that the large-scale use of bioenergy, if not implemented properly, would raise food prices and increase the number of people at risk of hunger in many areas of the world. For example, an increase in global bioenergy demand from 200 to 300 EJ causes a − 11% to + 40% change in food crop prices and decreases food consumption from − 45 to − 2 kcal person−1 day−1, leading to an additional 0 to 25 million people at risk of hunger compared with the case of no bioenergy demand (90th percentile range across models). This risk does not rule out the intensive use of bioenergy but shows the importance of its careful implementation, potentially including regulations that protect cropland for food production or for the use of bioenergy feedstock on land that is not competitive with food production.

Cloud and convective parameterizations strongly influence uncertainties in equilibrium climate sensitivity. We provide a proof‐of‐concept study to constrain these parameterizations in a perturbed parameter ensemble of the atmosphere‐only version of the Goddard Institute for Space Studies Model E2.1 simulations by evaluating model biases in the present‐day runs using multiple satellite climatologies and by comparing simulated δ18O of precipitation (δ18Op), known to be sensitive to parameterization schemes, with a global database of speleothem δ18O records covering the Last Glacial Maximum (LGM), mid‐Holocene (MH) and pre‐industrial (PI) periods. Relative to modern interannual variability, paleoclimate simulations show greater sensitivity to parameter changes, allowing for an evaluation of model uncertainties over a broader range of climate forcing and the identification of parts of the world that are parameter sensitive. Certain simulations reproduced absolute δ18Op values across all time periods, along with LGM and MH δ18Op anomalies relative to the PI, better than the default parameterization. No single set of parameterizations worked well in all climate states, likely due to the non‐stationarity of cloud feedbacks under varying boundary conditions. Future work that involves varying multiple parameter sets simultaneously with coupled ocean feedbacks will likely provide improved constraints on cloud and convective parameterizations. Plain Language Summary Equilibrium climate sensitivity (ECS) is a key climate metric that quantifies the rise in global mean surface temperature in response to doubling of atmospheric CO2. Changes in hydroclimate, temperature extremes, and other aspects of future climate projections are closely tied to a model's ECS. For decades, ECS range has remained wide despite improvements from using multiple lines of evidence. One persistent source of this spread is related to cloud and convective processes, which occur at scales too small to be explicitly resolved, and thus require parameterizations to be represented in climate models. These parameterizations directly influence water isotopes by modulating simulated clouds and atmospheric circulation, and thus can be used to constrain model processes and identify model biases. In this work, we demonstrated that paleoclimate simulations are more parameter sensitive than the modern, highlighting the potential of past climates in discriminating cloud and convective parameterizations. Using satellite‐ and proxy‐model comparisons, we identified the top performing parameterizations which differ for each time period likely due to varying cloud feedbacks under diverse climatic forcing. Overall, our results provide a framework for fine‐tuning model representations using combined paleoclimate and satellite data, offering a unique opportunity to assess model uncertainties over a broader range of climate variability. Key Points Paleoclimate relative to modern are more parameter sensitive, allowing for an assessment of uncertainties over a variety of climate forcing Certain simulations reproduced the δ18O of precipitation from paleoclimate proxies better than the default parameterization No single set of parameters works well in all climate states likely due to varying boundary conditions influencing cloud feedbacks

Satellite observations of column‐averaged water isotopes are relatively new retrieval products that are in need of further in situ evaluation. Such evaluation studies are generally difficult to perform due to the wide mismatch in temporal and spatial scales between the satellite observations based on instantaneous pixel averages during an overpass and airborne in situ measurements ranging up to several hours over a km‐scale. In addition, topography, weather conditions and in particular cloudiness impose severe constraints on an exact collocation between satellite and airborne in situ measurement platforms. Here we present a new method that allows a comparison between in situ measurements and satellite observations of δD on a broader statistical basis. We use regional isotope‐enabled model simulations as intermediate information to identify the area for best comparisons. Applying our methodology to TROPOMI total column δD retrievals for the L‐WAIVE campaign in Annecy, France, during June 2019 increases the number of satellite pixels for comparison despite widespread cloudiness on average by a factor of 20. In addition, the comparison of simulated and observed δD revealed a dependency of the satellite evaluation on the structure of the middle and upper troposphere. We conclude that our method provides a more robust statistic basis for in situ evaluation of δD satellite retrievals. The method will thus be useful in planning and executing forthcoming validation and evaluation campaigns, and can potentially be used for the evaluation of other satellite products. Plain Language Summary A characteristic of atmospheric water vapor is the concentration of stable heavy hydrogen and oxygen isotopes. Isotopic concentrations are observed by diverse techniques such as remote sensing from satellites and land‐based instruments, and direct measurements from aircrafts, ships and stations on land. These measurements lead to a variety of data sets which span different distances (from a few to hundreds of km) and time periods (from seconds to days). While remote measurements provide data of a large spatial and temporal coverage at a coarse resolution, direct measurements are often obtained during research campaigns over a limited time period with high spatial and temporal resolution. Especially the later data sets are of high value as they describe the atmospheric state in high detail. In this study we develop a method to extrapolate direct, campaign based high‐quality measurements of water vapor to the largest possible representative area using output from a numerical weather simulation. This allows us to compare remotely sensed and direct isotope measurements. This comparison illustrates how to interpret total column measurement and to identify where the model has difficulties to correctly simulate the vertical isotope distribution. This knowledge is of use for future application of remotely sensed data sets and model development. Key Points New, expanded collocation criterion provides larger data sets for comparison of δD observations during a field campaign We evaluate TROPOMI δD column retrievals with in situ airborne measurements using a COSMOiso‐based collocation criterion The combination of subcolumn and total column averaged δD is used to characterize vertical δD gradients and reveal model biases

Featured paper: See also the Editorial by McDowell et al

Paleoclimate observations constitute the only constraint on climate behavior prior to the instrumental era. However, such observations only provide indirect (proxy) constraints on physical variables. Proxy system models aim to improve the interpretation of such observations and better quantify their inherent uncertainties. However, existing models are currently scattered in the literature, making their integration difficult. Here, we present a comprehensive modeling framework for proxy systems, named PRYSM. For this initial iteration, we focus on water‐isotope based climate proxies in ice cores, corals, tree ring cellulose, and speleothem calcite. We review modeling approaches for each proxy class, and pair them with an isotope‐enabled climate simulation to illustrate the new scientific insights that may be gained from this framework. Applications include parameter sensitivity analysis, the quantification of archive‐specific processes on the recorded climate signal, and the quantification of how chronological uncertainties affect signal detection, demonstrating the utility of PRYSM for a broad array of climate studies. Key Points: A new modeling framework for paleoclimate proxies is proposed (PRYSM) PRYSM bridges the gap between GCMs and paleoclimate observations PRYSM may improve interpretation and uncertainty quantification of paleodata

Bayesian approaches to the modelling of ecological systems are increasingly popular, but there are competing methods for formal model comparisons. Here, we focus on the task of performing multimodel inference through estimating posterior model weights, which encompasses uncertainties in the choice of competing model structure into the inference outputs. Model‐based approaches such as reversible‐jump Markov chain Monte Carlo (RJ‐MCMC) are flexible and allow multimodel inference, but can be complex to implement and optimise, and so we translate a model‐based approach for ecological applications using Importance Sampling to estimate the marginal likelihood of the data given a particular model. This approach allows for model comparison through the estimation of Bayes' Factors or interpretable posterior model probabilities, yielding model weights that facilitate multimodel inference through Bayesian model averaging. We demonstrate Importance Sampling with two case study investigations in animal demography: censused analysis of banded mongoose (Mungos mungo) survival where missing data are uncommon, and capture–mark–recapture analysis of European badger (Meles meles) survival where data are commonly missing. We compare outcomes of the model comparison using the Importance Sampling approach to those obtained through single‐model inference approaches using Deviance information criteria and the Watanabe–Akaike information criteria. The results of the Importance Sampling method aligns with RJ‐MCMC model comparisons while often being more straightforward to fit and optimise, particularly if the competing models are non‐nested.

The Department of Energy’s (DOE’s) National Energy Technology Laboratory’s (NETL’s) Blowout and Spill Occurrence Model (BLOSOM), and the National Oceanic and Atmospheric Administration’s (NOAA’s) General NOAA Operational Modeling Environment (GNOME) are compared. Increasingly complex simulations are used to assess similarities and differences between the two models’ components. The simulations presented here are forced by ocean currents from a Finite Volume Community Ocean Model (FVCOM) implementation that has excellent skill in representing tidal motion, and with observed wind data that compensates for a coarse vertical ocean model resolution. The comprehensive comparison between GNOME and BLOSOM presented here, should aid modelers in interpreting their results. Beyond many similarities, aspects where both models are distinct are highlighted. Some suggestions for improvement are included, e.g., the inclusion of temporal interpolation of the forcing fields (BLOSOM) or the inclusion of a deflection angle option when parameterizing wind-driven processes (GNOME). Overall, GNOME and BLOSOM perform similarly, and are found to be complementary oil spill models. This paper also sheds light on what drove the historical Point Wells spill, and serves the additional purpose of being a learning resource for those interested in oil spill modeling. The increasingly complex approach used for the comparison is also used, in parallel, to illustrate the approach an oil spill modeler would typically follow when trying to hindcast or forecast an oil spill, including detailed technical information on basic aspects, like choosing a computational time step. We discuss our successful hindcast of the 2003 Point Wells oil spill that, to our knowledge, had remained unexplained. The oil spill models’ solutions are compared to the historical Point Wells’ oil trajectory, in time and space, as determined from overflight information. Our hindcast broadly replicates the correct locations at the correct times, using accurate tide and wind forcing. While the choice of wind coefficient we use is unconventional, a simplified analytic model supported by observations, suggests that it is justified under this study’s circumstances. We highlight some of the key oceanographic findings as they may relate to other oil spills, and to the regional oceanography of the Salish Sea, including recommendations for future studies.

Regeneration is an essential demographic step that affects plant population persistence, recovery after disturbances, and potential migration to track suitable climate conditions. Challenges of restoring big sagebrush (Artemisia tridentata) after disturbances including fire‐invasive annual grass interactions exemplify the need to understand the complex regeneration processes of this long‐lived, woody species that is widespread across the semiarid western U.S. Projected 21st century climate change is expected to increase drought risks and intensify restoration challenges. A detailed understanding of regeneration will be crucial for developing management frameworks for the big sagebrush region in the 21st century. Here, we used two complementary models to explore spatial and temporal relationships in the potential of big sagebrush regeneration representing (1) range‐wide big sagebrush regeneration responses in natural vegetation (process‐based model) and (2) big sagebrush restoration seeding outcomes following fire in the Great Basin and the Snake River Plains (regression‐based model). The process‐based model suggested substantial geographic variation in long‐term regeneration trajectories with central and northern areas of the big sagebrush region remaining climatically suitable, whereas marginal and southern areas are becoming less suitable. The regression‐based model suggested, however, that restoration seeding may become increasingly more difficult, illustrating the particularly difficult challenge of promoting sagebrush establishment after wildfire in invaded landscapes. These results suggest that sustaining big sagebrush on the landscape throughout the 21st century may climatically be feasible for many areas and that uncertainty about the long‐term sustainability of big sagebrush may be driven more by dynamics of biological invasions and wildfire than by uncertainty in climate change projections. Divergent projections of the two models under 21st century climate conditions encourage further study to evaluate potential benefits of re‐creating conditions of uninvaded, unburned natural big sagebrush vegetation for post‐fire restoration seeding, such as seeding in multiple years and, for at least much of the northern Great Basin and Snake River Plains, the control of the fire‐invasive annual grass cycle.